Reading in the data

I joined two datasets together. The main dataset I use came from mrpantherson on Kaggle and the other came from Gabrielle Baldassarre on Kaggle.

library(RSQLite)
library(tidyverse)
library(lubridate)
library(plotly)
library(pander)

fivethousand <- read_csv("C:/Users/james/Documents/git/M335_FA18_Sant_Keat/data/bgg_db_2018_01.csv")
filename <- "C:/Users/james/Documents/git/M335_FA18_Sant_Keat/data/database.sqlite"
sqlite.driver <- dbDriver("SQLite")
db <- dbConnect(sqlite.driver,
                dbname = filename)

#dbListTables(db)
boardgames <- dbReadTable(db,"BoardGames")

Wrangling the data

The structure for the data was not normalized and I found a few columns in which multiple values were placed in the same cell. For example, all the publishers for a game were found in one cell for each game. This would be a problem if I wanted to individually show data on each publisher. Using regular expression, I separated the values and placed each publisher in their own cell.

publisher <- boardgames %>% 
  select(details.name, attributes.boardgamepublisher) %>% 
  rename(names = details.name, publisher = attributes.boardgamepublisher)
  
# join sql publisher dataset with dataset of 5000 rows
games <- fivethousand %>% 
  left_join(publisher, by="names") %>% 
  filter(year > 0) %>% 
  mutate(year = as.integer(year))

# Found pattern and seperated by common character ","
games_p <- games %>%
  {mutate(.,publisher = str_replace_all(publisher,
                                        ",(?=[^\\s])","/"))} 

games_publisher <- games_p %>% 
  separate(publisher, into = paste0(rep("p",67), 1:67), sep = "/") %>% 
  gather(key = "p_id", value = "publisher", 21:87) %>% 
  filter(publisher != 0, !is.na(publisher)) %>% 
  select(-category, -designer, -mechanic, -bgg_url, -image_url, -p_id)

find_pub <- function(pub) {
  games_publisher %>% 
    rename(Complexity = weight, 
           Rating = geek_rating,
           Game = names,
           Time = max_time,
           Players = max_players) %>% 
    filter(publisher == pub, 
           year >= 1990, Time < 400, 
           Players <= 30) %>% 
    ggplot(aes(Complexity, Rating, size = Players, color = Time, group = Game)) +
    geom_point() +
    scale_x_continuous(limits = c(1,5)) +
    labs(title = paste("Board Games", pub, "Helped Publish Since 1990"), color = "Playing Time \n(minutes)", y = "Rating (1-10)", x = "Complexity (1-5)") +
    theme_minimal()
}

Research

As I studied the data I noticed that there were board games recorded 3000 B.C. I also noted that there were board games that required thousands of minutes and about 99 people to play. Because the general population is not interested in those types of board games, I filtered them from the dataset to focus more on the board games we are interested in playing. I took all the board games that were published after 1950 until the beginning of 2018 and created a histogram to show the progression of board games being published over the years. The graph below shows that board games consistantly increased in popularity. There was a huge boom for board games starting in the 90’s and continued to climb all the way until 2015.

I researched what happened in 1990 to have such an affect on board games and I found one of the connections was the internet. The internet brought new ways for publishers to run kickstarters and raise money used to produce board games and it gave them away to easily contact fans to test their board games out before release. I found my research at Financial Review. Also the rise of video games gave board games a boost. It gave an idea that games were not just for children but that adults could also play them.

games_publisher %>% 
  select(names, year) %>% 
  distinct() %>% 
  filter(year > 1950) %>% 
  ggplot(aes(year)) +
  geom_bar(color = "dodgerblue", fill = "dodgerblue4") +
  labs(title = "Quantity of Board Games Published Since 1950", y = "Number of Games Published", caption = "Fig. 1") +
  theme_minimal() +
  theme(axis.title.x = element_blank())

Although there are a lot of really great games, I wanted to focus on the main publishers. Publishers that publish hundreds of games abviously make a lot of good games, otherwise they wouldn’t sell any and they wouldn’t be able to make more. They thrive off the satisfactory of their clients. So I selected the top 10 publishers to see how many board games they publish per year and to see the quality of their games. The two plots below represent this. In Fig. 2 the publishers are in order from most board games published to least board games published.

games_publisher %>% 
  filter(year >= 1990, year != 2018) %>% 
  mutate(year2015 = year == 2015) %>%
  group_by(publisher) %>% 
  mutate(count = n()) %>% 
  ungroup() %>% 
  filter(count >= 167) %>% 
  mutate(publisher = fct_reorder(publisher, count, max, .desc = T)) %>% 
  ggplot(aes(year, fill = year2015)) +
  geom_bar() +
  scale_x_continuous(limits = c(1990, 2020), breaks = seq(1990, 2020, 5)) +
  scale_y_continuous(breaks = seq(0, 30, 15)) +
  scale_fill_manual(values = c("dimgrey", "green")) +
  facet_wrap(~publisher, nrow = 10, strip.position = "right") +
 labs(title = "Top 10 Board Game Publishers Since 1990", subtitle = "2015 had the most board games published in one year", y = "Number of Boardgames Published", x = "Year", caption = "Fig. 2") +
  theme_minimal() +
  theme(strip.text.y = element_text(angle = 0, hjust = 0),
        panel.grid.major = element_blank(),
        panel.grid.minor.x = element_blank(),
        axis.text.y = element_blank(),
        plot.background = element_rect(fill = "beige"),
        panel.grid.minor.y = element_line(color = "grey80"),
        axis.line.x = element_line(),
        axis.ticks.x.bottom = element_line(),
        legend.position = "none",
        axis.title.x = element_blank())

Fig.3 Shows a strong positive correlation between complexity and the rating, but it is interesting to see that the correlation can be interrupted by the length of the game. GMT Games have a lot of board games that take hours to play and the rating decreases the longer it takes to learn to play them.

games_publisher %>% 
  filter(year >= 1990) %>% 
  group_by(publisher) %>% 
  mutate(count = n()) %>% 
  ungroup() %>% 
  filter(count >= 167, max_time < 400, max_players <= 30) %>%
  ggplot(aes(weight, geek_rating, color = max_time)) +
  geom_point() +
  scale_color_gradient(low = "red", high = "yellow", aesthetics = "color") +
  scale_alpha(range = c(.3, .95)) +
  scale_x_continuous(limits = c(0,5)) +
  facet_wrap(~publisher, ncol = 5) +
  labs(title = "Quantity and Rating of Board Games Published By The Top 10 Publishers Since 1990", x = "Learning Complexity (1-5)", y = "Rating (1-10)", color = "Playing time \n(minutes)", alpha = "Maximum \nPlayers", caption = "Fig. 3") +
  theme_minimal() +
  theme(panel.grid.major = element_blank(),
        axis.title.y = element_text(vjust = 3.5),
        axis.title.x = element_text(vjust = -.2),
        panel.spacing = unit(.5, "lines"))

newdf <- games_publisher %>% 
  rename(Publisher = publisher) %>%
  filter(year >= 1990) %>% 
  group_by(Publisher) %>% 
  mutate(Published = n(), `Average Rating` = mean(geek_rating)) %>% 
  ungroup() %>% 
  filter(Published >= 167, max_time < 400, max_players <= 30) 
  
top_10_pub <- newdf %>% 
  select(Publisher, Published, `Average Rating`) %>% 
  distinct() %>% 
  arrange(desc(`Average Rating`))
         
pander(top_10_pub)